Automated Classification and Cluster Visualization of Genotypes Derived from High Resolution Melt Curves.

Area of Science:

• Molecular Biology
• Bioinformatics
• Genetics

Background:

• High Resolution Melting (HRM) analysis is a PCR-based method for DNA genotype identification.
• Current software for HRM genotype distinction is not fully automated and requires analyst interaction.
• Existing tools are limited to research applications and lack complete automation.

Purpose of the Study:

• To develop a fully automated machine learning algorithm for DNA genotyping using HRM.
• To classify unknown DNA genotypes accurately without user interpretation.
• To provide a robust and efficient method for genetic analysis.

Main Methods:

• A machine learning algorithm was developed to classify unknown genotypes from HRM data.
• Dynamic melt curves were transformed into multidimensional clusters using a training set.
• Probabilistic and statistical methods were employed for genotype classification on VKORC1, CYP2C9, and MTHFR assays.

Main Results:

• The automated algorithm achieved 100% accuracy for VKORC1, CYP2C9*3, and MTHFR c.665C>T genotyping.
• 97.5% of CYP2C9*2 samples were genotyped correctly, with 2.5% classified as 'no call'.
• The system demonstrated high accuracy and systematic classification of genotypes from HRM curves.

Conclusions:

• Full automation of DNA genotyping from HRM curves is achievable with high accuracy.
• The developed software eliminates the need for user interpretation in genetic analysis.
• The system offers visualization of genotype clusters and misclassification rates for assay optimization.